NFV (Network Function Virtualization) aims to implement some network functions on a general high-performance server, switch, and storage by using an IT virtualization technology. The network functions need to be implemented by using software and be capable of running on general server hardware. Moreover, an operation such as instantiation, auto scaling, or migration needs to be automatically performed on the network functions according to a requirement.
In an existing NFV environment, VNF (virtualized network function) running information may be obtained in two manners.
In a first manner, VNF service running information, that is, internally recorded VNF running information, is obtained by using an EMS (element management system). The VNF service running information includes fault information, performance information, and the like (for example, whether software runs normally and a data packet processing rate).
In a second manner, running information of a bottom-layer virtual resource, that is, running information of a resource used by a VNF, is obtained by using a VIM (virtualized infrastructure manager). The running information similarly includes fault information, performance information, and the like (for example, CPU usage and link bandwidth usage).
However, in an existing network, not all network elements have an EMS. For example, a value-added service device, represented by a firewall, a video acceleration device, or the like, has no dedicated EMS for management of the device. Therefore, when this type of service is deployed in the NFV environment, VNF service running information cannot be obtained by using the EMS. However, using only the running information obtained by using the VIM is insufficient to determine a VNF service status completely and in a timely manner. For example, when link bandwidth usage suddenly increases/decreases, MANO (management and orchestration) in NFV cannot determine a specific fault cause, but can only perform limited processing, for example, active/standby switchover, and reporting to an OSS (operations support system) in NFV to request further fault diagnosis. This is likely to increase troubleshooting time, and therefore affects quality of network service provisioning.